ghostvvriter

Content creation engine for Visualize Value. An MCP server that generates articles, drafts tweets, and creates visual concepts from Jack Butcher's archive.

No searching. No setup. Connect it and say "write me an article."

Install

npx ghostvvriter

Claude Code

claude mcp add ghostvvriter -- npx ghostvvriter

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "ghostvvriter": {
      "command": "npx",
      "args": ["ghostvvriter"]
    }
  }
}

Tools

Create

| Tool | What it does | |------|-------------| | suggest_article | Surfaces article ideas from the tweet archive and visual library. No input needed — every call shuffles the sample. | | generate_article | Writes the article from selected tweets and visuals. Output is publish-ready HTML with embedded tweets and images. | | draft_tweet | Drafts tweets in the VV voice. Optional topic. | | visual_ideas | Generates concepts for new VV-style illustrations. Optional theme. | | apply_framework | Applies a VV framework (productization spectrum, shu-ha-ri, time ladder, etc.) to any situation. |

Reference

| Tool | What it does | |------|-------------| | get_writing_profile | Jack Butcher's complete writing profile — voice, patterns, mechanics. | | search_tweets | Search the tweet archive by keyword. | | top_tweets | Top-performing tweets by likes. | | search_visuals | Search the visual library by text, tags, or description. | | get_visual | Get a specific visual by ID. | | get_daily_visual | Today's VV visual. | | list_frameworks | List all VV frameworks. | | get_framework | Get a specific framework. | | list_courses | List all VV courses (all free). | | get_course | Get a course curriculum. | | get_lesson | Get full lesson content by title. | | search_lessons | Search across all lesson content. | | list_projects | List all VV art projects. | | get_project | Get project details. | | search_projects | Search projects by keyword. |

How it works

The archive is the fuel, not the product. suggest_article loads a randomized sample of ~250 tweets (top performers, mid-tier, deep cuts) alongside ~150 VV visuals, and the AI finds the interesting idea clusters. Every call shuffles the sample, so you never get the same suggestions twice.

The creation tools require no input by default. The archive surfaces the ideas — you don't need to know what to look for.

Built by

Visualize Value · Jack Butcher

License

MIT